CN102081706A - Process planning method based on similarity theory - Google Patents

Abstract

A process planning method based on similarity theory comprises the following steps of: i) completing a part characteristic information model by adding a characteristic combination description on the basis of traditional single characteristic description, and building an description of part characteristics; ii) indicating entities formed by relevant information of relevant characteristic processing procedures regarding the characteristics as core by process elements, and generating a most basic module and a procedure characteristic of process planning; iii) building a case indexing structure based on the characteristics, generating similar process information by the similarity judgment of a part characteristic multi-branched tree, and performing the process similarity judgment; iv) performing procedure combinations by an agglomerative hierarchical clustering method, so as to generate a procedure cluster; and v) sequencing all the process elements in the procedure cluster according to the precedence relation, so as to ensure that all the process elements in the procedure cluster can be processed on a machine tool.

Description

Process planning method based on similarity theory

Technical field

What the present invention relates to is the method in a kind of process planning field, particularly a kind of process planning method based on similarity theory.

Background technology

The manufacturing system of can recombinating is to be driven by customer demand and manufacturing environment, has dynamic reorganization ability, and can respond the extraneous manufacturing system that changes fast by the Adjustment System configuration, and its maximum characteristics are to carry out self reorganization according to the variation in the external world.The reorganization of manufacturing system relates to system layer and two levels of mechanical floor, how system layer distributes the manufacturing resource rationally according to production task if mainly solving, and guarantee that streams, information flow are seamless mobile between each subsystem, thereby realize production function and productive capacity according to production task customization manufacturing system, this is to realize by the configuration of changing system.Therefore, system layer relates generally to following problems: a series of problems such as the recombinated layout of system-level process planning, manufacturing system and the manufacturing system logistic optmum of can recombinating.

The reorganization of manufacturing system is that product and technique change thereof drive, and the planning of the manufacturing system of can recombinating is based upon on the process planning basis.Therefore, process planning is one of the important component part of manufacturing system of can recombinating, but traditional process planning towards be manufacturing environment and the certain products object of fixing, and be not suitable for resource and processing object all is the manufacturing environment of dynamic change.Therefore, the maximum feature of process planning is its flexible and dynamic response under the manufacturing environment of can recombinating, it can make the process program of the dynamic change generation optimization of resource according to product and enterprise, and can in time handle the workshop accident, and makes corresponding technology adjustment.

Find that through literature search for the process planning under the manufacturing environment of can recombinating, A.Azab etc. propose the method for process planning and the integrated design of restructural lathe, realize the combination of operation in the process planning by the mapping of part feature and lathe module to prior art.The quantitative manufacturability appraisal procedure of application such as B.Lee proposes a kind of recombinated manufacturing workpiece family method for designing under given production schedule condition.Zhang Xiaofeng uses for reference the part thought of non-linear process planning, closed loop process planning and distributed process planning, has proposed the process planning system design framework towards the manufacturing cell that can recombinate.Ni Zhonghua etc. are on the basis of the characteristics of analyzing the manufacturing system of can recombinating and manufacturing system new requirement to process planning, propose a kind of process planning architecture and functional mode that satisfies the manufacturing system of to recombinate, and set up the recombinated process planning system that is applicable to machine-tool industry.

More than only on theoretical aspect, inquire into and study towards the process planning method of the manufacturing environment of can recombinating, lack practicality.And in the process planning process of the manufacturing system of can recombinating, it is exactly to reuse ripe process program that an important rationalization requires, with as far as possible little modification cost, design the technology that needs as soon as possible, thereby reduce design cost and production cycle.Between the structure example of part and the job operation, existing correspondence is the reusable basis of process program between analog structure and the similar technology.Therefore, be necessary to study the dynamic process planing method that to recombinate under the manufacturing environment based on similarity theory, and,, realize the reorganization of technological process by reusing corresponding process knowledge according to the status information of technological process in the process.

Summary of the invention

The objective of the invention is deficiency, propose a kind of process planning method based on similarity theory at existing process planning method, with many kinds, change under the satisfied manufacturing environment of can recombinating in batches, and the requirement of high product quality and agility.Make enterprise when part process is planned, can reuse the ready-made technology that had designed simultaneously, reduce design cost and production cycle.

The present invention realizes by following technical scheme:

A kind of process planning method based on similarity theory, it may further comprise the steps:

I) on the basis of the single feature description of tradition, add the characteristics combination description and improve the part feature information model, set up the description of part feature.

Ii) the formed entity of relevant information with the features relevant manufacturing procedure that is characterized as core is represented with process element, generated the basic module and the process feature of process planning.

Iii) set up an example index structure based on feature, the similarity determination by the part feature multiway tree generates similar technique information, carries out the resemble process sex determination.

Iv) use the hierarchical clustering method of cohesion and carry out the operation combination, generate operation family.

V) all process elements in the operation family are carried out the precedence relationship ordering, guarantee that all process elements in the operation family all can be processed on lathe.

Described parts information based on feature is described, and is meant the single feature description structure of the general employing of part feature description, from the angle of making, is rational with single feature description part.But this method is overemphasized the independence of feature, has ignored association and syntagmatic between the feature, causes that the process planning result's is inaccurate.

Describe and improve the part feature information model by on the basis of single feature description, adding characteristics combination.Characteristics combination is exactly according to the requirement of process planning to feature, gets up to describe with describing the characteristics combination that is relative to each other in single feature of part, and illustrates the character of this association.

The generation method of described process feature is meant: the formed entity of relevant information with the features relevant manufacturing procedure that is characterized as core is defined as process element, and is expressed as with five-tuple

pe _ij＝(d _i，g _i，b _i，MP _ij，MR _ij)i＝1，2，…，n；j＝1，2，…，m

Wherein, d _iBe feature f _iAffiliated azimuth plane; g _iBe feature f _iForm and position tolerance; b _iBe feature f _iWith the related classification between the further feature; MP _IjExpression is to feature f _iProcess the job operation that is adopted; MR _IjProcessing technology MP is adopted in expression _IjMachining feature f _iThe machining precision that is reached.Process element is the elementary cell of part process planning, is the basic module of forming technological process.

Whole process elements of a feature constitute the processing unit of this feature, and are expressed as

Pe _i＝{pe _i1，pe _i2，…，pe _im}

Process is a changing process of part feature configuration, because in process, each feature all is in a kind of machining state, so the status information of per pass operation demarcated by a plurality of feature process unit, and the set of these feature process units is called process feature.The form of embodying is:

P _t＝{pe _1，t，pe _2，t，…，pe _n，t}

Process feature is that the feature process unit with part is the center, and the process element in each stage shines upon one group of machining information.Be made up of a plurality of process feature from the blank to the part, each process feature all comprises corresponding process element and machining information.

Foundation is based on the example index structure of feature, the factors such as accuracy class, roughness and form and position tolerance with type, material type, heat treatment method, blank type and the shape facility of the part type relevant with similarity coefficient, processing mode, feature that are meant are divided into part level, shape level and feature level, set up a tree type example index structure based on feature.This structure can be abstract is the multiway tree structure, and the node in the tree is represented an object, multiway tree visual representation the mutual relationship between object, this hierarchical relationship has just been represented the similarity between object.

Resemble process calculating is meant and adopts bottom-up strategy, calculates local similarity earlier, calculates the similarity of the overall situation again, successively upwards carries out, and tries to achieve the total similarity evaluation of part example at last.The Tversky method is all adopted in the summation of each layer, that is:

$S=\sqrt{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i{F}_i^2}$

Wherein, two nodes to be compared have n property value, w _iBe the weight coefficient of i attribute, F _iIt is the similarity function of i attribute.Specific as follows:

(a) similarity of calculated characteristics

The attribute of employing feature comes the similarity of calculated characteristics.When two features are compared, adopt multiway tree node similarity computing method to calculate.Attribute in the feature generally is a value type, should carry out normalized earlier to these attributes, and processing mode commonly used is the similarity value that is mapped as in the span.

(b) similarity of calculating shape

Calculate the similarity of part shape according to the similarity of feature.Basic calculation method remains multiway tree node similarity calculation method.Calculate the shape similarity of two parts, just need one by one the relatively feature of two parts.Two features to be compared may be one of following three kinds of situations: 1. two features are identical; 2. two features are mated; 3. two features are unmatched.Finish the shape similarity of calculating part after all characteristic similarities calculate again.

(c) calculate the part similarity

After obtaining the part shape similarity,, press the similarity that the Tversky method is calculated part in conjunction with other attribute of part level.

Based on the operation combination of cohesion level clustering procedure, it specifically comprises as follows:

(a) data normalization

If domain PE=is (pe ₁, pe ₂..., pe _v) be all process element set, wherein, v representation feature process element sum.Each process element pe _iAvailable 5 attributes are represented, promptly affiliated azimuth plane, form and position tolerance, related classification, job operation and accuracy class.So, obtain the raw data matrix PE of process element:

${\mathrm{PE}}_{v\×5}=\left[\begin{array}{ccccc}{\mathrm{pe}}_{\mathrm{1,1}}& {\mathrm{pe}}_{\mathrm{1,2}}& {\mathrm{pe}}_{\mathrm{1,3}}& {\mathrm{pe}}_{\mathrm{1,4}}& {\mathrm{pe}}_{\mathrm{1,5}}\\ {\mathrm{pe}}_{\mathrm{2,1}}& {\mathrm{pe}}_{\mathrm{2,2}}& {\mathrm{pe}}_{\mathrm{2,3}}& {\mathrm{pe}}_{\mathrm{2,4}}& {\mathrm{pe}}_{\mathrm{2,5}}\\ .& .& .& .& .\\ .& .& .& .& .\\ .& .& .& .& .\\ {\mathrm{pe}}_{v,1}& {\mathrm{pe}}_{v,2}& {\mathrm{pe}}_{v,3}& {\mathrm{pe}}_{v,4}& {\mathrm{pe}}_{v,5}\end{array}\right]$

Because the dimension and the order of magnitude of 5 attributes are all inequality, have unified comparability in order to make ATTRIBUTE INDEX, must carry out dimensionless to each desired value and handle, and data compression is arrived on interval [0,1].Method commonly used has the standard deviation conversion, promptly

${\mathrm{pe}}_{i,k}^{\text{'}}=\frac{{\mathrm{pe}}_{i,k}-{\stackrel{\‾}{\mathrm{pe}}}_k}{s_k}(i=\mathrm{1,2},...,v;k=\mathrm{1,2,3,4,5})$

Wherein, ${\stackrel{\‾}{\mathrm{pe}}}_k=\frac{1}{v}\underset{i=1}{\overset{v}{\mathrm{\Σ}}}{\mathrm{pe}}_{i,k},s_k=\sqrt{\frac{1}{v}\underset{i=1}{\overset{v}{\mathrm{\Σ}}}{({\mathrm{pe}}_{\mathrm{ik}}-{\stackrel{\‾}{\mathrm{pe}}}_k)}^2}.$

(b) distance calculation between process element

Use d ₂(pe _i, pe _j) ∈ [O, 1] describes by cluster process element pe _iAnd pe _jBetween distance (distinctiveness ratio), adopt the Euclidean distance method to calculate d ₂(pe _i, pe _j).For ease of analyzing, further object distance unitization, adopt following European computing formula to carry out normalized, || pe _IkOne pe _Jk|| ₁Expression process element pe _iAnd pe _jDistance.

$d_2({\mathrm{pe}}_i,{\mathrm{pe}}_j)={\left|\right|{\mathrm{pe}}_i-{\mathrm{pe}}_j\left|\right|}_2={\left[\frac{1}{5}\underset{k=1}{\overset{5}{\mathrm{\Σ}}}{{\left|\right|{\mathrm{pe}}_{\mathrm{ik}}-{\mathrm{pe}}_{\mathrm{jk}}\left|\right|}_1}^2\right]}^{1/2}$

(c) distance calculation between operation family

Adopt different distance criterions can obtain different hierarchy clustering methods, minimum distance criterion is represented distance between family by the minimum a pair of data point of distance between family between family, makes in the family process element number too much, and similarity is lower, and the cluster effect is undesirable; Adopt the ultimate range criterion, the process element similarity is higher in the family, but the process element number is less, and operation family is more, and the cluster effect is also undesirable.Therefore, adopt the mean distance method to be calculated as follows:

$d_{\mathrm{avg}}(C_i,C_j)=\frac{1}{v_i{v}_j}\underset{\mathrm{pe}\∈C_i}{\mathrm{\Σ}}\underset{{\mathrm{pe}}^{\text{'}}\∈C_j}{\mathrm{\Σ}}\left|\right|\mathrm{pe}-{\mathrm{pe}}^{\text{'}}\left|\right|$

Wherein, d _Avg(C _i, C _j) the expression C of family _iWith the C of family _jBetween distance, || pe-pe ' || the distance of indicated object pe and object pe '.

(d) operation dynamic clustering

By (b), (c), calculate distinctiveness ratio matrix corresponding to the raw data matrix of process element:

According to the every distance value that provides in the distinctiveness ratio matrix, at first v process element is divided into v different family respectively, adopt the mean distance method to calculate distance between family, determine a threshold value, merging into a new operation family apart from family smaller or equal to threshold values.

The ordering of process element, according to the ordering rule between the process element in related classification that may occur between single feature and the operation family precedence relationship between the process element in the operation family is sorted, make that all process elements can machine successively in the operation family on equipment.Comprise:

(a) if between the feature benchmark relation is arranged, before then will being placed on, after feature has the process element of benchmark relation to be placed on therewith as the process element of benchmark.

(b) if between the feature benchmark relation is each other arranged, then according to the sequence arrangement process element order of feature priority as benchmark.

(c) if between the feature positioning relation is arranged, will determine that then the process element of further feature position is placed on before, determined that the process element of position is placed on after.

(d) if between the feature secondary relationship is arranged, then the sequencing that is produced by feature is determined the sequencing of process element.

(e) if the position accuracy demand relation is arranged between the feature, then these process elements are processed in the one-time positioning clamping.

(f) if similar feature association relation is arranged between the feature, then these process elements are concentrated in the least possible operation.

Owing to adopted above technical scheme, the invention has the beneficial effects as follows: can adapt to the requirement of modern manufacturing industry rapidly, adapt to the market demand of many kinds, short run to enterprise product quality and agility.By recycling enterprise existing ready-made process planning, and technology is made up, thereby shorten the process-cycle and reduce manufacturing cost according to demands such as part quality and throughput rate.

Description of drawings

Fig. 1 is the process planning process flow diagram based on similarity theory.

Fig. 2 is that operation combination of the present invention and operation family divide synoptic diagram.

Embodiment

The present invention is further illustrated below in conjunction with content of the present invention and the described embodiment of accompanying drawing.

A kind of process planning method of the present invention based on similarity theory, it may further comprise the steps:

Step 1 is described based on the parts information of feature.

Be used in and add the characteristic information model that characteristics combination describes on the basis of single feature description part feature information is described.According to the requirement of process planning, get up to describe with describing the characteristics combination that is relative to each other in single feature of part, and illustrate the character of this association feature.The related classification that may occur between single feature can be divided into following six kinds of relations according to technological requirement: 1. benchmark relation; 2. benchmark relation each other; 3. positioning relation; 4. secondary relationship; 5. accuracy requirement concerns; 6. similar feature association relation.

Step 2, process feature generates.

The formed entity of relevant information with the features relevant manufacturing procedure that is characterized as core is defined as process element, and is expressed as with five-tuple

pe _ij＝(d _i，g _i，b _i，MP _ij，MR _ij)i＝1，2，…，n；j＝1，2，…，m

Wherein, d _iBe feature f _iAffiliated azimuth plane; g _iBe feature f _iForm and position tolerance; b _iBe feature f _iWith the related classification between the further feature; MP _IjExpression is to feature f _iProcess the job operation that is adopted; MR _IjProcessing technology MP is adopted in expression _IjMachining feature f _iThe machining precision that is reached.Process element is the elementary cell of part process planning, is the basic module of forming technological process.

Whole process elements of a feature constitute the processing unit of this feature, and are expressed as

Pe _i＝{pe _i1，pe _i2，…，pe _im}

Process is a changing process of part feature configuration, because in process, each feature all is in a kind of machining state, so the status information of per pass operation demarcated by a plurality of feature process unit, and the set of these feature process units is called process feature.The form of embodying is:

P _t＝{pe _1，t，pe _2，t，…，pe _n，t}

With the core of process feature as the part feature process element, the process element in each stage shines upon one group of machining information.Be made up of a plurality of process feature from the blank to the part, each process feature all comprises corresponding process element and machining information.

Step 3, the resemble process sex determination.

The factors such as accuracy class, roughness and form and position tolerance of type, material type, heat treatment method, blank type and the shape facility of the part type relevant with similarity coefficient, processing mode, feature are divided into part level, shape level and feature level, set up a tree type example index structure based on feature.This structure can be abstract is the multiway tree structure, and the node in the tree is represented an object, the mutual relationship between multiway tree visual representation object.

Bottom-up strategy is adopted in resemble process calculating, calculates local similarity earlier, calculates the similarity of the overall situation again, successively upwards carries out, and tries to achieve the total similarity evaluation of part example at last.The Tversky method is all adopted in the summation of each layer, that is:

$S=\sqrt{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i{F}_i^2}$

Wherein, two nodes to be compared have n property value, w _iBe the weight coefficient of i attribute, F _iIt is the similarity function of i attribute.

(a) similarity of calculated characteristics

The attribute of employing feature comes the similarity of calculated characteristics.When two features are compared, adopt multiway tree node similarity computing method to calculate.Attribute in the feature generally is a value type, should carry out normalized earlier to these attributes, and processing mode commonly used is the similarity value that is mapped as in the span.

(b) similarity of calculating shape

Calculate the similarity of part shape according to the similarity of feature.Basic calculation method remains multiway tree node similarity calculation method.Calculate the shape similarity of two parts, just need one by one the relatively feature of two parts.Two features to be compared may be one of following three kinds of situations: 1. two features are identical; 2. two features are mated; 3. two features are unmatched.Finish the shape similarity of calculating part after all characteristic similarities calculate again.

(c) calculate the part similarity

After obtaining the part shape similarity, in conjunction with the similarity of other property calculation part of part level.

Set the similarity coefficient threshold value, when the similarity of part when setting threshold values, think and find similar part.Part and example are divided into two classes to the feature of part and example part through after the similar coupling: the feature of coupling and the feature of not mating.The technique information that keeps matching characteristic is deleted unnecessary feature process information in the example part, can get the process feature of formed technique information of new parts matching characteristic process element and per pass operation thus.To the feature of not mating, add the processing metamessage of this Partial Feature.

Step 4, process combination.

Employing is made up operation based on cohesion level clustering procedure, and its concrete steps are as follows:

(a) data normalization

If domain PE=is (pe ₁, pe ₂..., pe _v) be all process element set, wherein, v representation feature process element sum.Each process element pe _iAvailable 5 attributes are represented, promptly affiliated azimuth plane, form and position tolerance, related classification, job operation and accuracy class.So, obtain the raw data matrix PE of process element:

${\mathrm{PE}}_{v\&times;5}=\left[\begin{array}{ccccc}{\mathrm{<figure-callout\; id="1"\; label="pe"\; filenames="HDA0000048142890000021.png"\; state="\{\{state\}\}">\; <figure-callout\; id="1"\; label="pe"\; filenames="HDA0000048142890000021.png"\; state="\{\{state\}\}">pe</figure-callout>\; </figure-callout>}}_{\mathrm{1,1}}& {\mathrm{<figure-callout\; id="1"\; label="pe"\; filenames="HDA0000048142890000021.png"\; state="\{\{state\}\}">pe</figure-callout>}}_{\mathrm{1,2}}& {\mathrm{<figure-callout\; id="1"\; label="pe"\; filenames="HDA0000048142890000021.png"\; state="\{\{state\}\}">pe</figure-callout>}}_{\mathrm{1,3}}& {\mathrm{<figure-callout\; id="1"\; label="pe"\; filenames="HDA0000048142890000021.png"\; state="\{\{state\}\}">\; <figure-callout\; id="4"\; label="pe"\; filenames="HDA0000048142890000011.png"\; state="\{\{state\}\}">pe</figure-callout>\; </figure-callout>}}_{\mathrm{1,4}}& {\mathrm{<figure-callout\; id="1"\; label="pe"\; filenames="HDA0000048142890000021.png"\; state="\{\{state\}\}">pe</figure-callout>}}_{\mathrm{1,5}}\\ {\mathrm{<figure-callout\; id="1"\; label="pe"\; filenames="HDA0000048142890000021.png"\; state="\{\{state\}\}">pe</figure-callout>}}_{\mathrm{2,1}}& {\mathrm{pe}}_{\mathrm{2,2}}& {\mathrm{pe}}_{\mathrm{2,3}}& {\mathrm{<figure-callout\; id="4"\; label="pe"\; filenames="HDA0000048142890000011.png"\; state="\{\{state\}\}">pe</figure-callout>}}_{\mathrm{2,4}}& {\mathrm{pe}}_{\mathrm{2,5}}\\ .& .& .& .& .\\ .& .& .& .& .\\ .& .& .& .& .\\ {\mathrm{pe}}_{v,1}& {\mathrm{pe}}_{v,2}& {\mathrm{pe}}_{v,3}& {\mathrm{pe}}_{v,4}& {\mathrm{pe}}_{v,5}\end{array}\right]$

Because the dimension and the order of magnitude of 5 attributes are all inequality, have unified comparability in order to make ATTRIBUTE INDEX, must carry out dimensionless to each desired value and handle, and data compression is arrived on interval [0,1].Method commonly used has the standard deviation conversion, promptly

${\mathrm{pe}}_{i,k}^{\text{'}}=\frac{{\mathrm{pe}}_{i,k}-{\stackrel{\&OverBar;}{\mathrm{pe}}}_k}{s_k}(i=\mathrm{1,2},...,v;k=\mathrm{1,2,3,4,5})$

Wherein, ${\stackrel{\&OverBar;}{\mathrm{pe}}}_k=\frac{1}{v}\underset{i=1}{\overset{v}{\mathrm{\&Sigma;}}}{\mathrm{pe}}_{i,k},s_k=\sqrt{\frac{1}{v}\underset{i=1}{\overset{v}{\mathrm{\&Sigma;}}}{({\mathrm{pe}}_{\mathrm{ik}}-{\stackrel{\&OverBar;}{\mathrm{pe}}}_k)}^2}.$

(b) distance calculation between process element

Use d ₂(pe _i, pe _j) ∈ [O, 1] describes by cluster process element pe _iAnd pe _jBetween distance (distinctiveness ratio), adopt the Euclidean distance method to calculate d ₂(pe _i, pe _j).For ease of analyzing, further object distance unitization, adopt following European computing formula to carry out normalized, || pe _IkOne pe _Jk|| ₁Expression process element pe _iAnd pe _jDistance.

$d_2({\mathrm{pe}}_i,{\mathrm{pe}}_j)={\left|\right|{\mathrm{pe}}_i-{\mathrm{pe}}_j\left|\right|}_2={\left[\frac{1}{5}\underset{k=1}{\overset{5}{\mathrm{\&Sigma;}}}{{\left|\right|{\mathrm{pe}}_{\mathrm{ik}}-{\mathrm{pe}}_{\mathrm{jk}}\left|\right|}_1}^2\right]}^{1/2}$

(c) distance calculation between operation family

Adopt different distance criterions can obtain different hierarchy clustering methods, minimum distance criterion is represented distance between family by the minimum a pair of data point of distance between family between family, makes in the family process element number too much, and similarity is lower, and the cluster effect is undesirable; Adopt the ultimate range criterion, the process element similarity is higher in the family, but the process element number is less, and operation family is more, and the cluster effect is also undesirable.Therefore, adopt the mean distance method to be calculated as follows:

$d_{\mathrm{avg}}(C_i,C_j)=\frac{1}{v_i{v}_j}\underset{\mathrm{pe}\&Element;C_i}{\mathrm{\&Sigma;}}\underset{{\mathrm{pe}}^{\text{'}}\&Element;C_j}{\mathrm{\&Sigma;}}\left|\right|\mathrm{pe}-{\mathrm{pe}}^{\text{'}}\left|\right|$

Wherein, d _Avg(C _i, C _j) be the C of family _iWith the C of family _jBetween distance, || pe-pe ' || the distance of indicated object pe and object pe '.

(d) operation dynamic clustering

By (b), (c), calculate distinctiveness ratio matrix corresponding to the raw data matrix of process element:

According to the every distance value that provides in the distinctiveness ratio matrix, at first v process element is divided into v different family respectively, adopt the mean distance method to calculate distance between family, determine a threshold value, the family that distance is not more than threshold values merges into a new operation family.

Step 5, the process element ordering.

According to the ordering rule between the process element in related classification that may occur between single feature and the operation family precedence relationship between the process element in the operation family is sorted, make that all process elements can machine successively in the operation family on equipment.

Flow process of the present invention as shown in Figure 1.Mainly comprise several parts such as feature description 2, characteristic information 3, example index 4, similarity coupling 5, processing unit 6, operation combination 7, process element ordering 8, process establishment and output 9, example organization and management 10, case library 11, example storage 12, feature database 13, process element 14, process feature 15, combination principle 16 and lathe storehouse 17.Its detailed step is as follows:

(1) earlier preprocessing part 1 is carried out signature analysis, utilize all machining feature of 2 pairs of parts 1 of character description method that single feature and assemblage characteristic combine to be described.

(2) repetitive operation (1), up to new parts 1 all machining feature are described, and with all feature description generating feature information 3.

(3) characteristic informations 3 such as accuracy class, roughness and form and position tolerance with type, material type, heat treatment method, blank type and the shape facility of the part type relevant with similarity coefficient, processing mode, feature are divided into part level, shape level and feature level, set up a tree type example index structure 4 based on feature.

(4), adopt multiway tree node similarity and Tversky method and in conjunction with machining feature property calculation characteristic similarity according to the example index structure 4 of the new parts 1 set up.

(5) calculate the similarity of part shape according to the similarity of feature.Basic calculation method remains multiway tree node similarity calculation method.Calculate the shape similarity of two parts, just need one by one the relatively feature of two parts.Finish the shape similarity of calculating part after all characteristic similarities calculate again.

(6) after the acquisition part shape similarity, in conjunction with other attribute of part level, the Tversky method is calculated the similarity 5 of part.

(7) part and example are divided into two classes to the feature of part and example part through after the similar coupling: the feature of coupling and the feature of not mating.The technique information 15 that keeps matching characteristic is deleted unnecessary feature process information in the example part, can get the process feature of formed technique information of new parts matching characteristic process element and per pass operation thus.

(8) for the feature of coupling not in the operation (7), at first forming process element 14, then with the set generative process feature 15 of these process elements with the mapping by feature database 13 of the relevant manufacturing procedure information that is characterized as core.

(9) utilize the technique information 6 and the process feature 15 to keeping of cohesion level clustering procedure to carry out the operation combination.The process that operation combination and operation family divide has been described to Fig. 2 imagery.

(10) according to the ordering rule between the process element in related classification that may occur between single feature and the operation family precedence relationship between the process element in the operation family in the operation (9) is sorted, realize operation ordering 8, make that all process elements can machine successively in the operation family on equipment.

(11) process establishment and output and editor that output 9 realizes technology will store in the craft embodiment storehouse 11 through example storage 12 through the new parts technology of confirming, and manage by 10 pairs of case libraries of example organization and management.

The above-mentioned description to embodiment is can understand and apply the invention for ease of those skilled in the art.The person skilled in the art obviously can easily make various modifications to these embodiment, and needn't pass through performing creative labour being applied in the General Principle of this explanation among other embodiment.Therefore, the invention is not restricted to the example here, those skilled in the art should be within protection scope of the present invention for improvement and modification that the present invention makes according to announcement of the present invention.

Claims (10)

1. process planning method based on similarity theory, it is characterized in that: it may further comprise the steps:

I) on the basis of the single feature description of tradition, add the characteristics combination description and improve the part feature information model, set up the description of part feature;

Ii) the formed entity of relevant information with the features relevant manufacturing procedure that is characterized as core is represented with process element, generated the basic module and the process feature of process planning;

Iii) set up an example index structure based on feature, the similarity determination by the part feature multiway tree generates similar technique information, carries out the resemble process sex determination;

Iv) use the hierarchical clustering method of cohesion and carry out the operation combination, generate operation family;

V) all process elements in the operation family are carried out the precedence relationship ordering, guarantee that all process elements in the operation family all can be processed on lathe.

2. method according to claim 1 is characterized in that: described characteristics combination is according to the requirement of process planning to feature, gets up to describe with describing the characteristics combination that is relative to each other in single feature of part, and illustrates the character of this association.

3. method according to claim 1, it is characterized in that: described process element is meant the formed entity of relevant information with the features relevant manufacturing procedure that is characterized as core, being the elementary cell of part process planning, is the basic module of forming technological process, and is expressed as with five-tuple

pe _ij＝(d _i，g _i，b _i，MP _ij，MR _ij)i＝1，2，…，n；j＝1，2，…，m

Wherein, d _iBe feature f _iAffiliated azimuth plane; g _iBe feature f _iForm and position tolerance; b _iBe feature f _iWith the related classification between the further feature; MP _IjExpression is to feature f _iProcess the job operation that is adopted; MR _IjJob operation MP is adopted in expression _IjMachining feature f _iThe machining precision that is reached.

4. method according to claim 1 is characterized in that: described process feature is that the feature process unit with part is the center, and the process element in each stage shines upon one group of machining information, and the form of embodying is:

P _t＝{pe _1，t，pe _2，t，…，pe _n，t}。

5. method according to claim 1, it is characterized in that: described foundation is based on the example index structure of feature, be meant the part type relevant with similarity coefficient, processing mode, the type of feature, material type, heat treatment method, the accuracy class of blank type and shape facility, factor such as roughness and form and position tolerance is divided into the part level, shape level and feature level, set up a tree type example index structure based on feature, this structure can abstractly be the multiway tree structure, node in the tree is represented an object, multiway tree visual representation the mutual relationship between object, this hierarchical relationship has just been represented the similarity between object.

6. method according to claim 5, it is characterized in that: described similarity coefficient is meant a parameter that is used to weigh new parts and example part likeness in form degree, it is not only relevant with factors such as part type, processing mode, characteristic types, and is also relevant with the factors such as accuracy class, roughness and form and position tolerance of material type, heat treatment method, blank type and the shape facility of part.

7. method according to claim 1 is characterized in that: described resemble process sex determination is meant and adopts bottom-up strategy, calculates local similarity earlier, calculates the similarity of the overall situation again, successively upwards carries out, and tries to achieve the total similarity evaluation of part at last; Wherein, the Tversky method is all adopted in the summation of each layer, that is:

$S=\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i{F}_i^2}$

Wherein, two nodes to be compared have n property value, w _iBe the weight coefficient of i attribute, F _iIt is the similarity function of i attribute.

8. method according to claim 6 is characterized in that: described resemble process sex determination comprises:

(a) similarity of calculated characteristics

The attribute of employing feature comes the similarity of calculated characteristics, when two features are compared, adopts multiway tree node similarity computing method to calculate;

(b) similarity of calculating shape

Calculate the similarity of part shape according to the similarity of feature, computing method are multiway tree node similarity calculation method; Wherein calculating the shape similarity of two parts, is the feature that compares two parts one by one, and comprise three kinds of situations: 1. two features are identical; 2. two features are mated; 3. two features are unmatched;

(c) similarity of calculating part

After obtaining the part shape similarity, in conjunction with attributes such as the material type of part level, heat treatment method, blank type, production type, the similarity of press Tversky method calculating part.

9. method according to claim 1 is characterized in that: the operation combination of the hierarchical clustering method of described application cohesion, and it specifically comprises:

(a) data normalization

If domain PE=is (pe ₁, pe ₂..., pe _v) be all process element set, wherein, v representation feature process element sum, each process element pe _iRepresent with 5 attributes, promptly under azimuth plane, form and position tolerance, related classification, job operation and accuracy class, obtain the raw data matrix PE of process element:

${\mathrm{PE}}_{v\&times;5}=\left[\begin{array}{ccccc}{\mathrm{pe}}_{\mathrm{1,1}}& {\mathrm{pe}}_{\mathrm{1,2}}& {\mathrm{pe}}_{\mathrm{1,3}}& {\mathrm{pe}}_{\mathrm{1,4}}& {\mathrm{pe}}_{\mathrm{1,5}}\\ {\mathrm{pe}}_{\mathrm{2,1}}& {\mathrm{pe}}_{\mathrm{2,2}}& {\mathrm{pe}}_{\mathrm{2,3}}& {\mathrm{pe}}_{\mathrm{2,4}}& {\mathrm{pe}}_{\mathrm{2,5}}\\ .& .& .& .& .\\ .& .& .& .& .\\ .& .& .& .& .\\ {\mathrm{pe}}_{v,1}& {\mathrm{pe}}_{v,2}& {\mathrm{pe}}_{v,3}& {\mathrm{pe}}_{v,4}& {\mathrm{pe}}_{v,5}\end{array}\right]$

Each desired value is carried out dimensionless handle, and data compression is arrived on interval [0,1], promptly

${\mathrm{pe}}_{i,k}^{\text{'}}=\frac{{\mathrm{pe}}_{i,k}-{\stackrel{\&OverBar;}{\mathrm{pe}}}_k}{s_k}(i=\mathrm{1,2},...,v;k=\mathrm{1,2,3,4,5})$

Wherein, ${\stackrel{\&OverBar;}{\mathrm{pe}}}_k=\frac{1}{v}\underset{i=1}{\overset{v}{\mathrm{\&Sigma;}}}{\mathrm{pe}}_{i,k},s_k=\sqrt{\frac{1}{v}\underset{i=1}{\overset{v}{\mathrm{\&Sigma;}}}{({\mathrm{pe}}_{\mathrm{ik}}-{\stackrel{\&OverBar;}{\mathrm{pe}}}_k)}^2};$

(b) distance calculation between process element

Use d ₂(pe _i, pe _j) ∈ [O, 1] describes by cluster process element pe _iAnd pe _jBetween distance, adopt the Euclidean distance method to calculate d ₂(pe _i, pe _j), and, adopt following European computing formula to carry out normalized with parasangization, || pe _IkOne pe _Jk|| ₁Expression process element pe _iAnd pe _jDistance,

$d_2({\mathrm{pe}}_i,{\mathrm{pe}}_j)={\left|\right|{\mathrm{pe}}_i-{\mathrm{pe}}_j\left|\right|}_2={\left[\frac{1}{5}\underset{k=1}{\overset{5}{\mathrm{\&Sigma;}}}{{\left|\right|{\mathrm{pe}}_{\mathrm{ik}}-{\mathrm{pe}}_{\mathrm{jk}}\left|\right|}_1}^2\right]}^{1/2};$

(c) distance calculation between operation family

Adopt the mean distance method to be calculated as follows:

$d_{\mathrm{avg}}(C_i,C_j)=\frac{1}{v_i{v}_j}\underset{\mathrm{pe}\&Element;C_i}{\mathrm{\&Sigma;}}\underset{{\mathrm{pe}}^{\text{'}}\&Element;C_j}{\mathrm{\&Sigma;}}\left|\right|\mathrm{pe}-{\mathrm{pe}}^{\text{'}}\left|\right|$

Wherein, d _Avg(C _i, C _j) the expression C of family _iWith the C of family _jBetween distance, || pe-pe ' || the distance of indicated object pe and object pe ';

(d) operation dynamic clustering

By (b), (c), calculate distinctiveness ratio matrix corresponding to the raw data matrix of process element:

According to the every distance value that provides in the distinctiveness ratio matrix, v process element is divided into v different family respectively, adopt the mean distance method to calculate distance between family, determine a threshold value, merging into a new operation family apart from family smaller or equal to threshold values.

10. method according to claim 1 is characterized in that: the ordering of described process element comprises:

(a) if between the feature benchmark relation is arranged, before then will being placed on, after feature has the process element of benchmark relation to be placed on therewith as the process element of benchmark;

(b) if between the feature benchmark relation is each other arranged, then according to the sequence arrangement process element order of feature priority as benchmark;

(c) if between the feature positioning relation is arranged, will determine that then the process element of further feature position is placed on before, determined that the process element of position is placed on after;

(d) if between the feature secondary relationship is arranged, then the sequencing that is produced by feature is determined the sequencing of process element;

(e) if the position accuracy demand relation is arranged between the feature, then these process elements are processed in the one-time positioning clamping;

(f) if similar feature association relation is arranged between the feature, then these process elements are concentrated in the least possible operation.

Images